Violence Against Civilians in the Second Intifada

Abstract

Patterns of military violence against civilians vary considerably, between conflicts and within them. This article explores why some combatants are more likely than others to engage in violence against civilians, focusing on a particular subset of such violence, violence that is not planned or authorized by military superiors, termed opportunistic violence. In a sample of Israeli combat soldiers from the Second Intifada, opportunistic violence was found to be strongly associated with duration of deployment among civilians. Long deployments erode social and moral norms, raising the likelihood that combatants will act opportunistically. However, the relationship between deployment duration and opportunistic violence was moderated by unit structure, such that long deployments were more likely to cause opportunistic violence in units with weak command structures. These findings suggest that the deleterious effects of long deployments are not inevitable but can be limited to a large extent through well-functioning structures of command and discipline.

Keywords

political violence Israel West Bank Gaza conflict

Violence against civilians in conflict varies considerably (Kalyvas, 2006; Weinstein, 2007), not only across wars (Downes, 2008; Valentino, Huth, & Balch-Lindsay, 2004) but also within them (Kalyvas, 2006). Violent behavior varies among individual combatants and armed groups, with some combat units exhibiting far higher levels of violence than others (Humphreys & Weinstein 2006). Given such variation in wartime violence patterns, an immediate question arises: What factors cause violence to decrease? Understanding these factors is of great practical importance, as a better understanding of what drives violence can help design policies that minimize wartime violence against civilians.

In this article, I explore a subset of military violence against civilians, violence that is not planned or ordered by superiors, termed opportunistic violence. Most empirical studies of violence against civilians have focused largely on strategic violence—violence that is authorized from above, explicitly or implicitly. In contrast, I focus on violence that is not in the strategic interest of military leaders, either for normative or utilitarian reasons. Such violence nevertheless occurs due to individual combatants who use violence for private gains (Humphreys & Weinstein, 2006) or have divergent norms (Wood, 2009). I define opportunistic violence as intentional harm inflicted by combatants on civilians that is (a) not planned, ordered, or authorized by military superiors and is (b) carried out by combatants for their own benefit rather than the benefit of the military organization. The distinction between strategic and opportunistic violence recognizes that militaries are not unitary actors but complex organizations with a range of actors, interests, and conflicting pressures. While military authorities, the principals, have certain interests regarding the form and scale of violence that should be used to achieve particular objectives, ordinary soldiers, or agents, are often motivated by very different goals (Mitchell, 2004; Weinstein, 2007; Wood, 2006). It is this form of agent-centered violence that is the focus of the present study.

To date, research on opportunistic military violence has been limited, most likely because data on such violence are difficult to obtain. By its nature, opportunistic violence is often concealed from view. Nevertheless, such violence is an ever-present risk, particularly in irregular warfare where boundaries between soldiers and civilians are notoriously blurred and fighting takes place in civilian areas rather than on the conventional battlefield.

I examine a particular case of military counterinsurgency and occupation, the Second Intifada (2000-2005). In the absence of systematic data regarding opportunistic violence, I use an exploratory survey of men who conducted their regular service in combat units in the Israeli Defense Force (IDF) during the insurgency.¹ An exploratory design is appropriate in areas in which little systematic research exists. Because a random sample of combat soldiers was not possible in the Israeli–Palestinian context, and probably in most contexts of ongoing conflict, I utilized two sampling methods to maximize the diversity of the sample: chain referral methods (snowball sampling) using several initial starting points and targeted sampling of military veteran groups on the social networking site Facebook. The sample was indeed diverse and included men from nearly all IDF combat units and from a wide variety of backgrounds and geographical locations. The goal of this study is thus not to make inferences regarding general violence patterns in the Second Intifada but rather to suggest provocative tendencies grounded in evidence regarding the correlates of opportunistic violence.

What accounts for variation in opportunistic violence among combat units? I consider two factors often linked to violence against civilians in the Israeli–Palestinian context: Duration of deployment among civilians, and revenge, arguing that in the Israeli–Palestinian case, social-psychological processes of moral disengagement and routinization are more likely to be associated with opportunistic violence than heated passions such as the desire for vengeance. Next I consider the effects of several organizational variables on opportunistic violence: unit morale, unit structure, and sanction systems, hypothesizing that levels of opportunistic violence are lower in units with high morale and well-functioning structures of command and control. Moreover, I hypothesize that command and control structures moderate the effects of deployment duration, such that prolonged deployments should have a stronger effect on opportunistic violence in units with weak command structures than in units with well-functioning structures of command and control.

Of the variables examined, I find that deployment duration, not revenge, is associated with opportunistic violence in the Israeli–Palestinian case. As social-psychological theories would lead us to expect, prolonged exposure to civilians from an enemy out-group increases the likelihood of opportunistic violence against that group due to such processes as moral disengagement, routinization, and desensitization. Importantly, however, this relationship is moderated by unit structure, such that the effects of prolonged deployments are much stronger in units in which command structures are weak. Armed group structure thus serves as a buffer against the harmful effects of prolonged deployments among civilians.

The Second Intifada

The Second Intifada, also known as the Al-Aqsa Intifada (after the mosque where the violence first broke out), is one of the longest and bloodiest episodes of the Israeli–Palestinian conflict. The outbreak of violence signaled the imminent collapse of the Oslo Peace Process, which had attempted to bring a negotiated solution to the conflict. The Second Intifada pitted Palestinian insurgents from a variety of organizations against the Israeli military that fought to repress the insurgency. Both sides used violence strategically to advance their goals, shifting their repertoires of strategic violence in accordance with changing imperatives.² Palestinian organizations used a variety of violent tactics against Israeli soldiers and civilians, including gunfire attacks on roads and settlements, bombings, rocket attacks, and most notably, a suicide terror campaign in Israeli cities. Israeli counterinsurgency tactics included assassinations of militants, bombings of suspected militant installations, widespread arrests and detentions, house demolitions, and severe restrictions on movement, eventually leading to military reoccupation of the West Bank cities from which the IDF had withdrawn under the Oslo Accords, and, 3 years later, to unilateral withdrawal from the Gaza Strip.

Between 2000 and 2005, more than 3,000 Palestinians and more than 1,000 Israelis were killed as a result of the outbreak of violence. Civilians bore the brunt of much of the violence, comprising an estimated two thirds of casualties on the Israeli side and one half on the Palestinian side.³ However, the statistics on lethal force only partially reflect the extent of violence experienced by parties to the conflict. In the course of repression of the insurgency, Palestinian civilians experienced a steep increase in Israeli military regimentation and control, leading to the loss of freedom of movement, widespread destruction of property, an ongoing threat of detention and arrest, and physical force and verbal abuse at sites of friction with the Israeli military. Israeli civilians suffered a major blow to their sense of personal security, with cafes, buses, and shopping malls becoming sites of bloody terrorist attacks. The economy took a major hit, and, at the peak of the Intifada, civilian life grinded to a near halt.

As an asymmetric conflict taking place in civilian areas rather than on a conventional battlefield, the Second Intifada brought Israeli soldiers into close contact with Palestinian civilians at checkpoints, roadblocks, homes, and streets. This contact created incentives for opportunistic violence against civilians beyond the violence authorized by military superiors. Indeed, local and international nongovernmental organizations (NGOs) documented many instances of such violence, as did the military itself. For instance, an internal IDF study leaked to the press found that approximately 25% of soldiers had engaged in or witnessed abusive behavior at checkpoints (Yehoshua, 2007). Nevertheless, the violence was not uniform in scale, and varied among combatants. In the following sections, I explore the determinants of some of this variation.

Theory and Hypotheses

Defining Violence Against Civilians in Conflict: Strategy Versus Opportunism

In the growing literature exploring the microdynamics of violence against civilians in conflict, two primary theories have emerged. The predominant theory focuses on the shifting strategic benefits of such violence. For instance, violence can be used to generate compliance from a resistant population (Mitchell, 2004), to obtain information and collaboration (Kalyvas, 2006), to signal determination to continue fighting in the face of deteriorating prospects so as to improve bargaining leverage (Hultman, 2007), to reduce losses and bring about victory in prolonged wars of attrition (Downes, 2006), and to counter shifts in threat levels (Mitchell, 2004). A second theory highlights the role of the organizational characteristics of armed groups, arguing that violence is the product of disciplinary breakdowns in some armed units (Humphreys & Weinstein 2006), differing recruitment processes (Weinstein, 2007), or in the case of sexual violence, of the inability of armed group leaders who oppose sexual violence to enforce their preferences on ordinary combatants through strong hierarchies (Wood, 2009).

Typically, the two theories are seen as conflicting. Thus, Humphreys and Weinstein (2006) observe that organizational explanations pose “a challenge to the view that high levels of abuse and violence are observed where leaders retain tight control over an efficient killing machine that can be directed at will” (p. 444). However, upon closer examination it becomes clear that the two theories do not contradict but instead are concerned with different types of violence against civilians, both of which commonly occur in conflict. Theorists of violence-as-strategy predefine violence as violence committed in accordance with armed group leader interests, or “strategic violence” (e.g., Downes, 2006; Kalyvas, 2006). Theorists of violence-as-indiscipline measure violence committed without the permission of armed group leader orders, or “opportunistic violence” (e.g., Humphreys & Weinstein, 2006). While, without the appropriate data, these forms of violence may be difficult to distinguish, they nevertheless stem from different processes: Strategic violence results from strategic considerations at the elite level, while opportunistic violence reflects the preferences and constraints of ordinary combatants, insofar as these diverge from the preferences of the elites.⁴

Duration of Deployment Among Civilians

One factor I propose is associated with opportunistic violence is the psychological effect of prolonged deployments among civilians from another group, in many cases, a group identified with the enemy. Compounding the ordinary preference that people give their own group over out-groups is the conflict situation that creates sizable asymmetries of power between soldiers and civilians. Civilians are low-status opponents, who tend to elicit contempt and disgust (Fiske, Harris, & Cuddy, 2004). Individual norms against violence erode over time with repeated exposure to moral dilemmas, through well-documented processes of moral disengagement (Bandura, 1999), dehumanization (Huggins, Haritos-Fatouros, & Zimbardo, 2002), desensitization (Bandura, 1978), and routinization (Kelman & Hamilton, 1989). Importantly, the process that leads to violence against civilians is a gradual one. As Bandura (2002) observes,

Disengagement practices will not instantly transform considerate persons into cruel ones. Rather, the change is achieved by progressive disengagement of self-censure. Initially, individuals perform mildly harmful acts they can tolerate with some discomfort. After their self-reproof has been diminished through repeated enactments, the level of ruthlessness increases, until eventually acts originally regarded as abhorrent can be performed with little anguish or self-censure. (p. 110)

In the Israeli–Palestinian context, opportunistic violence is often attributed to prolonged deployment among civilians and to the resulting numbing and erosion of norms (e.g., Elizur & Yishay-Krien, 2009). Long-term deployments among civilians are frequently cited as explanations when cases of opportunistic violence surface in the Israeli media (e.g., Azulay, 2008; Blau, 2009; Harel, 2010). Accordingly, I raise the following hypothesis:

Hypothesis 1: Opportunistic violence increases in frequency as duration of deployment among civilians increases.

Revenge

An alternative theory attributes opportunistic violence against civilians to heated emotions and passions, most notably, the drive for revenge. Kalyvas (2006) argues that the motives of soldiers are often expressive, noting that “revenge is probably the most recurrent feature in the descriptions of violence in civil war” (p. 59). The approach linking violence against civilians to heated emotion is especially prevalent in the literature on ethnic conflict, much of which focuses on the role of intergroup animosities and passions in fostering violence (Horowitz, 1985; Kaufman, 2001, 2006; Petersen, 2002). The emphasis on heated emotions would seem to be particularly relevant to the Israeli–Palestinian case, as it is one of the paradigmatic examples of long-standing ethnic conflict. However, it stands in stark contrast to explanations such as Bandura’s (1999) theory of moral disengagement, which contend that, in many cases, violence results not from aggressive emotions and drives but from the gradual cognitive recasting of the meaning of violence so that it is no longer understood as wrong or immoral. The logic of emotion and revenge generates the following hypothesis:

Hypothesis 2: Opportunistic violence increases as the drive for revenge rises.

Organizational Factors

The characteristics of units to which combatants belong can have an important effect on the levels of opportunistic violence against civilians in conflict (Humphreys & Weinstein, 2006; Weinstein, 2007). State militaries in particular are totalizing institutions that attempt to control nearly every aspect of an individual member’s behavior, through such practices as uniform codes of appearance, 24-hour availability, subjection to military discipline and law, inability to negotiate working conditions, and long deployments (Moskos, 1977). Thus, factors operating at the level of the military unit should be expected to have an effect on violence. I explore two such factors here: unit morale, and command and discipline.

Unit Morale

Military morale has long been studied as a predictor of military effectiveness and performance (Britt & Dickinson, 2006; Gal & Manning, 1987). Morale is defined as “the enthusiasm and persistence with which a member of a group engages in the prescribed activities of that group” (Manning, 1991, p. 455). A high level of unit morale would thus indicate a large number of individuals with high enthusiasm. Unit morale has been linked to a number of positive outcomes, such as psychological well-being and resistance to stress (Britt, Dickinson, Moore, & Castro, 2007) and military effectiveness (Motowidlo & Borman, 1978). In light of the positive consequences of morale, it is plausible to hypothesize that units with high morale will be less likely to engage in self-serving, opportunistic behavior.

Hypothesis 3: Opportunistic violence is negatively related to high morale.

Command and Discipline

To enforce armed group leader notions of strategic violence and prevent opportunistic violence, the command hierarchy must function properly, allowing for communication along the chain and command (Wood, 2009). Small unit commanders must be strong and authoritative in the face of negative influences from below, and prohibitions against opportunistic violence must be inculcated in combatants through training, rules of engagement, and control systems. Accordingly, I examine three indicators of command and discipline: commander authority, compliance with operational rules, and sanction systems.

Commander authority

A strong command structure is necessary for commanders to exercise authority and for soldiers and commanders to communicate with one another. State militaries are typically centralized, hierarchical institutions. Nevertheless, as several scholars have observed (Gates, 2002; Mitchell, 2004; Wood, 2009), the military hierarchy is subject to principal–agent problems as commanders must enforce the orders of superiors down the chain of command. The strength of the military hierarchy will therefore depend on the strength of the individual links in the chain and their ability to enforce their authority on their subordinates. When commanders are weak and lack authority, competing sources of authority can emerge undermining the hierarchical group structure. In such an environment, it is hypothesized that opportunistic violence will be more prevalent, as commanders will be less able to enforce rules and norms prohibiting opportunistic violence against civilians.

Hypothesis 4: Opportunistic violence is more common in units where commanders are weaker.

Compliance with operational rules

In addition, soldiers in units with well-functioning command structures should exhibit compliance with operational rules, unrelated to the treatment of civilians. Commanders who are able to induce compliance with such rules as preparation of equipment, briefing and debriefing before and after military tasks, maintenance of alert conditions, and so on are likely to be stronger than those who do not or cannot enforce such rules. I therefore hypothesize that

Hypothesis 5: Units in which soldiers have a high degree of compliance with operational rules engage in less opportunistic violence.

Punishment of violence against civilians

Formal control systems are hypothesized to play a role in reducing levels of opportunistic violence. In units where punishment for opportunistic violence is severe, potential violators will be deterred from violent behavior. The logic of deterrence is not uncontested however. Empirical studies of the effects of organizational punishment have had mixed results, with some showing positive outcomes for punishment, others stressing the indirect costs of punishment, which are said to outweigh benefits, and still others finding a more complex, nonlinear relationship between severity of punishment and misbehavior (Tenbrunsel & Messick, 1999; Trevino, 1992; Trevino, Weaver, & Reynolds, 2006). Consistent with the more conventional deterrence model, I suggest that:

Hypothesis 6: Opportunistic violence is less prevalent in units where violence against civilians is punished severely.

Beyond its direct effect on opportunistic violence, I propose that unit control structures should moderate the detrimental effects of prolonged deployments on opportunistic violence. One of the responsibilities of commanders is to ensure soldiers remain in line with prescribed rules, regulations, and norms, even as the passage of time makes the preservation of such discipline increasingly difficult. Prolonged deployments among civilians should therefore have less effect on opportunistic violence in units in which control is strong, than in units in which control is weak. This logic generates two hypotheses:

Hypothesis 7: Commander authority acts as a moderating variable, reducing the association between duration of deployment among civilians and opportunistic violence.

Hypothesis 8: Severity of punishment acts as a moderating variable, reducing the association between duration of deployment among civilians and opportunistic violence.

Method

Data on opportunistic violence against civilians are notoriously difficult to obtain (Kalyvas, 2006). Human rights organizations, NGOs, and journalists typically gather information from victims, which can lead to a skewed understanding of violence. Perpetrator accounts, however, are hard to access during wartime. Combatants are often legally barred from speaking or fear reprisal if they do. Much of the available data on soldier behavior only come into being after a political transition is effected. Thus, scholars survey and interview ex-combatants in peacetime, and study the archives of military tribunals, truth commissions, and other investigative bodies, hoping to shed light on the motives of perpetrators. Yet such data are potentially subject to bias, as yesterday’s heroes attempt to protect themselves from today’s victors. Once their actions have been discredited, ex-combatants are likely to trivialize the extent of their participation, to blame others, or to deny outright the commission of violent acts (S. Cohen, 2001).

Israel presents a unique case in which military violence against civilians is highly scrutinized by domestic and international agencies. Nevertheless, there are as of yet no systematic data on the prevalence of opportunistic violence.⁵ In most cases, Palestinians do not file complaints against such violence, and due to its largely covert nature, it is difficult to systematically monitor through external observation. Some Israeli organizations collect testimonies from soldiers, but such soldiers are a self-selecting group that is widely perceived as unrepresentative of the soldier population. Systematic data on variation in opportunistic violence among combat units are nonexistent.⁶ While a few studies have used interview methods to investigate military violence in Israel (Elizur & Yishay-Krien, 2009; Ron, 2000a, 2003), this exploratory study benefits from unique survey data collected from ex-combatants in a context of ongoing conflict, thereby providing an important first step in the study of the correlates of opportunistic violence.

Procedure and Sample

Online surveys were circulated to veterans who conducted their regular service in the IDF between the years 1999 and 2006, explicitly asking respondents to address the period of their regular, as distinct from reserve, service. The time frame was limited so as to avoid inclusion in the sample of men who conducted their regular service during more recent episodes of the Palestinian–Israeli conflict and in particular, the 2009 Gaza War, which would have introduced additional complexity into the data set.

Former combatants constitute a hidden population, meaning that no publicly available sampling frame exists from which it could have been possible to select a random sample and that membership in the group raises issues of privacy or sensitivity (Heckathorn, 1997, 2002; Watters & Biernacki, 1989). Several sampling methods have been used to study hidden populations, including snowball sampling (N. Cohen & Arieli, 2011), targeted sampling (Watters & Biernacki, 1989), and location sampling. Each of these methods has deficiencies that have been widely documented. Nevertheless, they currently represent the only available sampling methods. While such methods are not representative by definition, it is possible to increase their quality through access to parallel networks and through the use of complementary methods.

To maximize the diversity of my sample I utilized two sampling methods. First I used chain referral sampling (snowball methods), aiming for diversity in selection of several initial respondents (seeds) from different units and social networks. A second method commonly used to study hidden populations is targeted location sampling, in which researchers recruit respondents at locations frequented by the relevant research population. Adapting this method to a location frequented by former soldiers, I replaced traditional location-based methods with “virtual” locations using the widespread social networking website Facebook. In Israel, Facebook use is especially prevalent, penetrating more than 40% of the population at the time the survey was conducted (Tsoref, 2010). In the age bracket I studied, Facebook has an even greater presence, cutting across gender, ethnicity, and class divides. I contacted former combatants through Facebook groups of military veterans in which they were members, enlisting the help of group administrators rather than contacting each member directly. Those administrators who agreed were asked to circulate the survey to group members.

In all, 195 individuals responded to the survey. Of those, 34 were disqualified as they did not fit the survey criteria. After deletion of those individuals who did not complete the survey or had substantial amounts of missing data, 118 surveys remained for analysis, 42% of which were sampled through Facebook and 58% through snowball sampling. Nearly all IDF regular ground units are represented in the survey, including its five infantry brigades, its armor, artillery and engineering corps, and its independent/special forces.

Measures

Opportunistic Violence

The dependent variable is an index summing four items containing descriptions of violence against civilians: physical violence, verbal abuse/humiliation, destruction of civilian property, and taking private property. While subject to important limitations, multiple-item self-report measures are a primary means of measurement in other domains of opportunistic behavior such as workplace deviance in organizational studies (Bennett & Robinson, 2000; Fox & Spector, 1999), and deviance and delinquency in criminology (Osgood, McMorris, & Potenza, 2002; Thornberry & Krohn, 2000), because in both fields, reliance on statistics of perpetrators who have been caught (such as supervisor reports or sentencing records) would likely vastly underestimate the problem.

The items in the scale were constructed based on extensive reviews of human rights literature on violence in the Israeli–Palestinian conflict, as well as dozens of interviews with former combatants, indicating that the four forms measured were primary forms of opportunistic violence in the Second Intifada. To ensure only opportunistic violence was measured and not violence ordered by military superiors, items were referred to in the survey as “infractions” and were included in a list of various infractions that sometimes occur in military units, the majority of which were not related to violence against civilians, such as refusing orders, destroying army property, violating guard duty, and so on.

To minimize the perceived threat of self-incrimination as well as to reduce social desirability bias, respondents were asked to report the frequency of each form of violence in their unit rather than their own engagement in violence. Responses were arranged on a Likert-type scale ranging from “nonexistent” (0) to “very high” (5). Interitem reliability analysis of the four items produced a relatively high Cronbach’s alpha of .85.⁷ As data on deviance and misconduct have the characteristic of being positively skewed, I transformed the dependent variable using a square root transformation before conducting analysis.

Duration of Deployment Among Civilians

Respondents were asked to indicate the areas in which their unit served, corresponding with the four major zones of activity for the combat units of the IDF: Israel’s northern border, Lebanon, the West Bank, and Gaza. They were also asked how long they were deployed in each area. Responses were measured on a 5-point scale indicating the following increments: “no time” (0), “less than 4 months” (1), “4 to 8 months” (2), “8 to 12 months” (3), “12 to 18 months” (4), and “more than 18 months” (5). Duration of service in the West Bank was used to measure deployment among civilians, as this was the zone in which most direct soldier–civilian interaction took place.⁸

Anger and Revenge

Two indicators were used to measure revenge. First, I collected data regarding the number of conflict-related combatant deaths per each IDF unit represented in the sample.⁹ Several studies have found battle losses to be related to strategic violence, both by states (Downes, 2006, 2008) and rebel groups (Hultman, 2007). By a similar logic, battle deaths might also drive opportunistic violence by creating rage and vengefulness among combatants. In the context of irregular war where distinctions between militants and civilians are blurred, the more deaths a particular unit sustains, the more likely it is to lash out in vengeance against civilians on the opposite side. Accounts of the My Lai massacre, for instance, emphasize the role of mounting casualties suffered by the Charlie Company in the weeks prior to the massacre, and the resulting demand for revenge among soldiers (Bilton & Sim, 1993; Olson & Roberts, 1998).

A second indicator of revenge is the number of civilians killed during a combatant’s period of service, averaged per month so as to enable comparison among combatants. The majority of Israeli casualties in the Second Intifada were civilians, many of whom were casualties of deadly suicide bombings. Such attacks were also likely to increase vengefulness among combatants, thereby contributing to opportunistic violence.

Unit Morale

Respondents were asked to rate the level of morale in their unit on a 5-point scale ranging from 1 (very low) to 5 (very high). This item is commonly used to measure morale in military surveys and in scholarly studies of military morale (e.g., Bliese & Britt, 2001; Maguen & Litz, 2006).

Command and Discipline

Three measures were utilized to measure command and discipline:

Commander authority

Commander authority was measured by the number of privileges exercised by the senior soldiers in the unit (i.e., those who have served in the unit for more than a year), on a scale ranging from 0 to 11. Units where senior soldiers received no extra privileges received a score of 11, and each privilege reduced the score by 1 point. From my qualitative research with Israeli soldiers, I learned that the number of privileges granted to the more senior soldiers is an important indicator of how well the unit’s hierarchy functions. Privileges, which are highly variable among units (see Table 1), range from the relatively benign (no more kitchen duties) to the disruptive (exemptions from disciplinary rules and requirements) to more extreme privileges that utterly undermine hierarchy, allowing senior soldiers to punish or impose tasks on their more junior counterparts. In units where senior soldiers are strong, commanders are by definition weak, as they are unable (or unwilling) to impose discipline on all but the most vulnerable members of the unit. A competing center of power is created within the unit, undermining the hierarchical structure. Indeed, for many new commanders, one of the first tasks in establishing authority is abolishing many of the privileges given to senior soldiers.

Table 1.

Summary Statistics.

Variable	n	M	SD	Minimum	Maximum
Opportunistic violence (square root)	118	1.75	1.13	0	4.47
Duration of deployment among civilians	118	2.13	1.74	0	5
Unit losses	118	4.16	3.33	0	14
Civilian deaths (per month of service)	118	9.00	5.86	0.44	23.37
Morale	118	3.79	0.86	1	5
Commander authority	118	7.87	2.22	2	11
Compliance	118	8.42	1.13	5	10
Punishment of opportunistic violence	84	2.15	0.69	0.25	3
Role	118	0.79	0.97	0	3
Elite unit	118	0.23	0.42	0	1

Compliance with operational rules

Respondents ranked on a 10-point scale to what extent seven operational rules were enforced in their unit, including briefing and debriefing before tasks, preparation of equipment, observation of alert conditions, and so on. Cronbach’s alpha for the seven items was .85.

Punishment of opportunistic violence

Soldiers were asked to identify the typical punishment a member in their unit would receive for each of the four items included in the dependent variable. Punishments were then ranked by severity as follows: “none” (0), “verbal reprimand” (1), “fine” or “grounding to base” (2), and “incarceration” (3). The punishment measure is an index of the average punishment for violence against civilians in the respondent’s unit. Cronbach’s alpha for the scale was .73.

Control Variables

I controlled for two additional variables that may have had effects on opportunistic violence: Whether respondents were soldiers, NCOs, or officers, and whether the respondent was in an elite unit. Elite units are special, voluntary IDF units that require soldiers to undergo a separate selection process to be admitted.

Results

Summary statistics for variables used in the analysis are provided in Table 1. While the dependent variable exhibits substantial variation, its mean is relatively low. This is likely attributable, at least in part, to the fact that opportunistic violence in an organized state military, like deviant behavior generally, is a relatively low base rate phenomenon. Nevertheless, it is of note that more than 80% of respondents reported having witnessed at least one incident of opportunistic violence in their unit, and approximately 70% reported two incidents or more. These figures are much higher than those found by the internal IDF study referenced earlier, which found 25% of respondents had witnessed or been involved in abusive behavior at checkpoints.¹⁰ Reliance on official statistics such as court or conviction records would have led to estimates that were much lower still (Yesh Din, 2008). The distribution of the opportunistic violence items is presented in Figure 1.

Figure 1.

Reported frequency of opportunistic violence in unit.

Table 2 reports the zero-order correlations among the dependent and independent variables.

Table 2.

Intercorrelations Among Study Variables.

Variable	2	3	4	5	6	7	8	9	10
1. Opportunistic violence (square root)	.31***	−.17	.15	.09	−.30***	−.27**	−.57***	−.003	−.24**
2. Duration of deployment among civilians		−.22**	−.10	.19*	−.04	.05	−0.08	−.12	.05
3. Unit losses			−.19*	.03	−.03	.13	.06	−.05	−.04
4. Civilian deaths per month				.02	−.15	.001	.07	−.16	−.05
5. Morale					.13	.21*	−.11	.03	.15
6. Commander authority						.01	.22*	.33***	.33***
7. Compliance with operational rules							.28*	−.04	.17
8. Punishment of opportunistic violence								.02	.16
9. Role									.08
10. Elite unit

p < .05. **p < .01. ***p < .001.

The results in Table 2, presenting bivariate relationships, indicate that duration of deployment among civilians is a significant predictor of opportunistic violence, in support of Hypothesis 1. The desire for revenge, as measured by unit losses and civilian deaths, is not associated with opportunistic violence and neither is morale, providing no support for Hypotheses 2 and 3. All three command and discipline measures are negatively and significantly associated with opportunistic violence, providing initial support for Hypotheses 4, 5, and 6. In addition, Table 2 shows that a number of the variables are correlated with one another, in particular, various measures used to measure command and discipline. Of the control variables, membership in an elite unit was negatively associated with opportunistic violence.

Standard ordinary least squares (OLS) regression was used to test the hypotheses and identify the independent effects of the measures used (Table 3).¹¹ In each of the multivariate models, I tested specifications with and without moderation effects. Because one of the measures—severity of punishment of opportunistic violence—contains many missing observations, I report in Columns 1 and 2 the results excluding the variable and Columns 3 and 4 the results including it. In Models 2 and 4, the predictor variables were centered before creating the interaction term (Aiken, West, & Reno, 1991).

Table 3.

Predictors of Opportunistic Violence—Multivariate Results.

Model	1	2	3	4
Duration of deployment among civilians	0.22***	0.21***	0.21***	0.21***
Unit losses	−0.02	−0.02	0.01	0.004
Civilian deaths	0.03	0.03	0.03	0.03
Morale	0.15	0.08	0.07	0.07
Commander authority	−0.14**	−0.16***	−0.13*	−0.12*
Compliance	−0.27**	−0.25**	−0.11	−0.12
Punishment of opportunistic violence			−0.76***	−0.75***
Role	0.17	0.17	0.17	0.16
Elite unit	−0.37	−0.22	−0.02	−0.03
Moderator variable: Duration of deployment among Civilians × Commander authority		−0.07**
Moderator variable: Duration of deployment among Civilians × Punishment of Opportunistic Violence				−0.09
Constant	3.83	3.00	4.24	1.78
Observations	118	118	84	84
R ²	.33	.39	.48	.49
Adjusted R²	.28	.33	.42	.42

p < .05. **p < .01. ***p < .001.

The effect of duration of deployment among civilians retains its size and significance in all models. In contrast, both measures of revenge have no significant relationship with opportunistic violence, and only the civilian deaths measure takes a positive sign. This provides strong support for the argument that disengagement and routinization, not rage and revenge, are associated with opportunistic violence in the Israeli–Palestinian context. As hypothesized, measures of command and control are negatively and significantly associated with opportunistic violence. Unit morale does not seem to be related to opportunistic violence, nor does soldier or commander status. Furthermore, once command and control variables are controlled for, the elite unit measure no longer exhibits independent effects. All of these findings are discussed in the next section.

Moderation Effects

As hypothesized, commander authority significantly moderated the relationship between deployment duration and opportunistic violence. However, while the interaction with severity of punishment carried the expected sign, it was not significant (p = .31), providing no support for Hypothesis 8.

Figure 2 shows the duration of deployment among civilians—opportunistic violence relationship for units with command structures that are one standard deviation above the mean (i.e., strong commanders) and units that are one standard deviation below the mean (i.e., weak commanders).

Figure 2.

The moderating effects of commander authority.

For soldiers in units with weak commanders, the simple slope for the relationship between duration of contact with civilians and opportunistic violence was positive and differed significantly from zero (b = .37, SE = .07, t = 5.1, p < .001). For soldiers who were in units with strong, authoritative commanders, the simple slope remained positive but did not differ significantly from zero (b = .05, SE = .08, t = 0.63, p = .53). This provides support for Hypothesis 7, as breakdown of commander authority strengthens the association between duration of deployment among civilians and opportunistic violence.

Discussion

The findings support the hypothesis that extended deployments among civilians, not the drive for revenge, are an important determinant of opportunistic violence. This finding is consistent with social-psychological theories regarding the gradual erosion of social and individual barriers against violence against a low-status out-group. While rage and revenge may play an important role in other contexts, the prolonged and routinized nature of Israeli counterinsurgency and military occupation in the Occupied Territories appears more likely to foster feelings of disengagement and desensitization than heated passions.

However, the findings also tell a more complex story about the deleterious effects of extended deployments, suggesting that these effects can be moderated by well-functioning command structures. The study thus provides additional support from the Israeli–Palestinian context to findings from Sierra Leone (Humphreys & Weinstein, 2006) and Sri Lanka (Wood, 2009) that emphasize the importance of armed group structure and discipline in reducing violence against civilians. In this section, I discuss some of the empirical findings, and then consider alternative explanations.

Structures of command and control, and specifically measures of hierarchy and discipline, are consistently found to have a significant effect on opportunistic violence. Furthermore, commander authority, though not punishment, was found to have a moderating effect, mitigating to a large extent the effects of long deployments among civilians on opportunistic violence. While the nonsignificant moderating role of punishment may simply be a result of the reduced sample size, it could also point to a substantive finding, suggesting that commanders may be more successful in inculcating prohibitions against opportunistic violence through such “positive” means as personal example and strong leadership, rather than through “negative” means such as severe punishment. Further research would be required to investigate this question.

Contrary to expectation, unit morale was not found to be significantly associated with opportunistic violence. A possible explanation for this is that opportunistic violence can itself raise levels of morale, creating bonds between participants through their witnessing of, or participation in, illicit behavior.¹² The data also suggest an interesting finding regarding elite units. In the bivariate analysis, elite units were found to be negatively and significantly associated with opportunistic violence, suggesting that members of elite units were less likely to engage in such behavior. This observed relationship is consistent with popular perceptions in Israel that admire members of elite units as “higher quality” soldiers. However, once factors related to command and control are controlled for, the association loses its significance, suggesting that the difference between elite and regular units is explained by the much stronger command and discipline structures existing in elite units rather than the individual characteristics of their members.

While the data provide initial support for theories highlighting the important role of hierarchy, discipline, and control in reducing violence, other explanations may be consistent with the observed relationships. One such explanation is that weak command and control structures and opportunistic violence may be typical of particular units that attract lower quality manpower. Indeed, some units in the IDF are less selective in the quality of manpower they recruit, a factor that may be associated with opportunistic violence and disciplinary breakdowns.

To assess the plausibility of this alternative explanation, I investigated the role of unit membership by creating dummy variables for each of the brigades included in the sample. A bivariate analysis of opportunistic violence and brigade membership produced a significant relationship in the case of only one unit, known today as the Kfir brigade. This finding is consistent with journalistic accounts, in which Kfir has been particularly notorious for opportunistic violence against civilians (e.g., Azulay, 2008; Fogelman, 2010; Harel, 2008). However, once brigade identity was entered into a multivariate regression controlling for duration of deployment and command and control factors, the effects of membership in Kfir were no longer significant. This suggests that long deployments and disciplinary problems in Kfir account for its relatively high share of opportunistic violence rather than any particular individual characteristics of its members. Indeed, Kfir is the only IDF brigade to serve solely in the West Bank, and its soldiers are deployed in one area for particularly long periods.¹³

A second explanation suggests that an alternative mechanism underlies the association between commander authority and opportunistic violence. It might be argued that the senior privilege measure is actually a measure of bad modeling rather than of commander authority. Social learning theory (Bandura, 1978) suggests that violence is learned through exposure to role models who model violent behavior as a method of conflict resolution or in response to stress. In units where senior soldiers mistreat junior soldiers, taking advantage of their power to haze, humiliate, punish, and impose ridiculous tasks, new recruits may learn to abuse their own power when faced with vulnerable civilians. However, the similar results found when compliance with operational rules and severity of punishment were used as measures lend additional support to the effects of command and control.

Limitations

This study is subject to some limitations in addition to the sampling issues discussed earlier. First, opportunistic violence is measured through self-reports, which raises issues of reconstructed memory (ex-combatants are asked to reflect on events that happened up to a decade earlier), social desirability, and fear of self-incrimination (Wood, 2003). It is also possible that those more guilty of opportunistic violence would be less inclined to participate in the survey in the first place, leading to an underestimation of the prevalence of violence. Attempts were made to minimize these risks through indirect questions and the anonymity and security of online surveys.

Second, though the sample is by definition not representative, I took several efforts to reduce bias and ensure diversity in my sample, using two sampling methods and targeting parallel networks. While the sample is indeed diverse in terms of units, zones of service, and demographic variables such as religious background and political attitudes, it does suggest some socioeconomic bias, with an overrepresentation of educated respondents from medium to high socioeconomic backgrounds. This is also reflected in the overrepresentation of commanders in the sample. Given the present impossibility of conducting a random sample, replication of the survey using different samples may be the only available method.

Third, this study focuses on situational and organizational variables, and does not investigate the effects of individual-level factors on opportunistic violence. Combatants come from different backgrounds and hold vastly different positions on the legitimacy of violence against civilians. Moreover, they differ in personalities, tendency to aggression, and ability to control impulses. All of these factors are likely to play some role in producing opportunistic violence (Fiske et al., 2004). In the Israeli–Palestinian context, Levy (2008) has argued that shifts in the demographic composition of the IDF have led to more aggressive military policies. Unfortunately the data did not allow me to test the effects of individual-level factors as the dependent variable measured violence at the unit level, while demographic information was available only for individual respondents.

Implications and Conclusion

In this article, I drew an analytic distinction between strategic and opportunistic violence, arguing that while both commonly occur in conflict, they result from different mechanisms, requiring disaggregated analysis. Noting the difficulties of collecting data on opportunistic violence, I conducted an exploratory study drawing on survey data from nearly 120 former IDF combatants, providing a unique opportunity to explore patterns of opportunistic violence in a context of ongoing conflict. This initial study demonstrates that survey studies on opportunistic violence in conflict, while challenging to run, are indeed feasible. Moreover, it demonstrates that self-report measures, though subject to important limitations, may reveal more about rates of opportunistic violence than alternative methods.

My findings suggest that extended deployments among civilians raise the likelihood of opportunistic violence as soldiers become more disengaged and desensitized over time. Together with mounting evidence regarding the negative effects of long deployments on the physical and mental health of soldiers (Adler, Huffman, Bliese, & Castro, 2005), this suggests that, where shortening deployments is not an option, soldiers and civilians may benefit from better care and support services to those deployed. Yet while long deployments are likely to have harmful effects on soldiers and civilians, the slide into violence is not inevitable and can be controlled at least to some extent. Units with well-functioning command structures are better able to delineate and enforce rules, ensuring that soldiers remain in line with the interests of superiors and refrain from opportunistic behavior. In units with malfunctioning command structures and lax discipline, violence against civilians is more likely to result.

This article has investigated the determinants of opportunistic violence in a particular context, that of the Second Intifada. However, it is plausible that a similar logic determining levels of violence holds in other state militaries engaged in prolonged counterinsurgency and occupation. In fact, the negative effects of long deployments among civilians may be even more acute in the case of other state militaries, because while Israeli soldiers who serve in the Occupied Territories regularly return home for brief periods, soldiers who are deployed far from home for months at a time are likely to be even more subject to the psychological mechanisms that gradually erode social and moral norms. Initial evidence for this proposition was provided by the 2007 Fifth Mental Health Advisory Team (MHAT V) study, commissioned by the Office of the U.S. Army Surgeon General to assess soldier health and well-being in Iraq. The study, which included a survey of 2,279 deployed soldiers, found deployment length to be one of the most significant risk factors for unethical behavior against noncombatants.¹⁴ Nevertheless, this article has argued that modern militaries that prize hierarchy, control, and discipline should be able to keep the troops in line, reigning in opportunistic violence at least to some extent.

On a final note, these findings call into question current prescriptions for dealing with the role of commanders in enabling civilian abuse, focusing primarily on the doctrine of command responsibility. The doctrine imposes criminal liability on a commander who knew or had reason to know of his subordinates’ misdeeds and did not take reasonable measures to prevent or punish such acts. Given growing evidence of the link between breakdowns in discipline and opportunistic violence, from settings as diverse as the United States, Sri Lanka, Israel, and Sierra Leone, it might well be argued that commanders bear some responsibility for the violence of their subordinates when they allow command and control structures to lapse, even in the absence of concrete or imputed knowledge regarding subordinate violence. The narrow lens of criminal liability may therefore be inadequate to capture the full extent of responsibility that commanders bear for the behavior of their troops during conflict.

Footnotes

Acknowledgements

I thank Edmond Keller, Elisabeth Wood, James Ron, Amit Gal, Robert Brym, Charli Carpenter, and three anonymous reviewers for valuable comments on earlier drafts.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This project has been supported by a Social Science Research Council International Dissertation Research Fellowship and a National Science Foundation Doctoral Dissertation Research Improvement Grant.

Notes

Author Biography

Devorah Manekin is a postdoctoral fellow at the Martin Buber Society of Fellows at the Hebrew University of Jerusalem, Israel. Her research interests include political violence, conflict processes, and international security.

References

Adler

Huffman

Bliese

P. D.

Castro

(2005). The impact of deployment length and experience on the well-being of male and female soldiers. Journal of Occupational Health Psychology, 10, 121-137.

Aiken

West

Reno

(1991). Multiple regression: Testing and interpreting interactions. Thousand Oaks, CA: Sage.

Azulay

(2008, February 24). Kfir soldiers abuse and officers admit: “We are in a bad period.” Haaretz (in Hebrew). Retrieved from http://www.haaretz.co.il/misc/1.1306754

Bandura

(1978). Social learning theory of aggression. Journal of Communication, 28(3), 12-29.

Bandura

(1999). Moral disengagement in the perpetration of inhumanities. Personality and Social Psychology Review, 3, 193-209.

Bandura

(2002). Selective moral disengagement in the exercise of moral agency. Journal of Moral Education, 31, 101-119.

Bennett

R. J.

Robinson

S. L.

(2000). Development of a measure of workplace deviance. Journal of Applied Psychology, 85, 349-360.

Bilton

Sim

(1993). Four hours in My Lai. New York, NY: Penguin.

Blau

(2009, April 1). Chief education officer in a letter to soldiers: Stop printing abusive statements on shirts. Haaretz (in Hebrew). Retrieved from http://www.haaretz.co.il/news/politics/1.1253525

10.

Bliese

P. D.

Britt

T. W.

(2001). Social support, group consensus and stressor–strain relationships: Social context matters. Journal of Organizational Behavior, 22, 425-436.

11.

Britt

T. W.

Dickinson

J. M.

(2006). Morale during military operations: A positive psychology approach. In Britt

T. W.

Adler

A. B.

Castro

C. A.

(Eds.), Military life: The psychology of serving in peace and combat (pp. 157-184). Westport, CT: Praeger Security International.

12.

Britt

T. W.

Dickinson

J. M.

Moore

Castro

(2007). Correlates and consequences of morale versus depression under stressful conditions. Journal of Occupational Health Psychology, 12, 34-47.

13.

Brym

R. J.

Araj

(2006). Suicide bombing as strategy and interaction: The case of the Second Intifada. Social Forces, 84, 1969-1986.

14.

Brym

R. J.

Maoz-Shai

(2009). Israeli state violence during the Second Intifada: Combining new institutionalist and rational choice approaches. Studies in Conflict & Terrorism, 32, 611.

15.

Cohen

Arieli

(2011). Field research in conflict environments: Methodological challenges and snowball sampling. Journal of Peace Research, 48, 423-435.

16.

Cohen

(2001). States of denial: Knowing about atrocities and suffering. Cambridge, UK: Cambridge University Press.

17.

Downes

A. B.

(2006). Desperate times, desperate measures: The causes of civilian victimization in war. International Security, 30, 152-195.

18.

Downes

A. B.

(2008). Targeting civilians in war. Ithaca, NY: Cornell University Press.

19.

Drucker

Shelah

(2005). Boomerang: Kishalon ha-manhigut ba-Intifadah ha-sheniya [Boomerang: The failure of leadership in the Second Intifada]. Jerusalem, Israel: Keter.

20.

Elizur

Yishay-Krien

(2009). Participation in atrocities among Israeli soldiers during the First Intifada: A qualitative analysis. Journal of Peace Research, 46, 251-267.

21.

Fiske

S. T.

Harris

L. T.

Cuddy

(2004). Why ordinary people torture enemy prisoners. Science, 306, 1482-1483.

22.

Fogelman

(2010, June 17). Negative contexts: Incidents involving the Kfir Brigade in the past three years. Haaretz. Retrieved from http://www.haaretz.com/weekend/magazine/negative-contexts-1.296823

23.

Fox

Spector

P. E.

(1999). A model of work frustration-aggression. Journal of Organizational Behavior, 20, 915-931.

24.

Gal

Manning

(1987). Morale and its components: A cross-national comparison. Journal of Applied Social Psychology, 17, 369-391.

25.

Gates

(2002). Recruitment and allegiance. Journal of Conflict Resolution, 46, 111-130.

26.

Harel

(2008, May 11). Kfir brigade leads in West Bank violations (in Hebrew). Haaretz. Retrieved from http://www.haaretz.co.il/news/politics/1.1226883

27.

Harel

(2010, October 26.) Officers testified in support of Givati soldiers convicted of endangering a child (in Hebrew). Haaretz.

28.

Heckathorn

(1997). Respondent-driven sampling: A new approach to the study of hidden populations. Social Problems, 44, 174-199.

29.

Heckathorn

(2002). Respondent-driven sampling II: Deriving valid population estimates from chain-referral samples of hidden populations. Social Problems, 49, 11-34.

30.

Horowitz

D. L.

(1985). Ethnic groups in conflict. Berkeley: University of California Press.

31.

Huggins

Haritos-Fatouros

Zimbardo

P. G.

(2002). Violence workers: Police torturers and murderers reconstruct Brazilian atrocities. Berkeley: University of California Press.

32.

Hultman

(2007). Battle losses and rebel violence: Raising the costs for fighting. Terrorism and Political Violence, 19, 205.

33.

Humphreys

Weinstein

(2006). Handling and manhandling civilians in civil war. American Political Science Review, 100, 429-447.

34.

Kalyvas

(2006). The logic of violence in civil war. Cambridge, UK: Cambridge University Press.

35.

Kaufman

S. J.

(2001). Modern hatreds: The symbolic politics of ethnic war. Ithaca, NY: Cornell University Press.

36.

Kaufman

S. J.

(2006). Symbolic politics or rational choice? Testing theories of extreme ethnic violence. International Security, 30(4), 45-86.

37.

Kelman

H. C.

Hamilton

V. L.

(1989). Crimes of obedience: Toward a social psychology of authority and responsibility. New Haven, CT: Yale University Press.

38.

Levy

(2008). The linkage between Israel’s military policies and the military’s social composition: The case of the al-Aqsa Intifada. American Behavioral Scientist, 51, 1575-1589.

39.

Maguen

Litz

(2006). Predictors of morale in U.S. peacekeepers. Journal of Applied Social Psychology, 36, 820-836.

40.

Manning

(1991). Morale, cohesion, and esprit de corps. In Gal

Mangelsdorff

A. D.

(Eds.), Handbook of military psychology (pp. 453-470). New York, NY: Wiley.

41.

Mitchell

N. J.

(2004). Agents of atrocity: Leaders, followers, and the violation of human rights in civil war. New York, NY: Palgrave-Macmillan.

42.

Moskos

(1977). From institution to occupation. Armed Forces & Society, 4, 41-50.

43.

Motowidlo

Borman

(1978). Relationships between military morale, motivation, satisfaction, and unit effectiveness. Journal of Applied Psychology, 63, 47-52.

44.

Olson

Roberts

(1998). My Lai: A brief history with documents. Boston, MA: Bedford Books.

45.

Osgood

Finken

McMorris

(2002). Analyzing multiple-item measures of crime and deviance II: Tobit Regression analysis of transformed scores. Journal of Quantitative Criminology, 18, 319-347.

46.

Osgood

McMorris

Potenza

(2002). Analyzing multiple-item measures of crime and deviance I: Item response theory scaling. Journal of Quantitative Criminology, 18, 267-296.

47.

Petersen

(2002). Understanding ethnic violence: Fear, hatred, and resentment in twentieth century Eastern Europe. Cambridge, UK: Cambridge University Press.

48.

Ron

(2000a). Savage restraint: Israel, Palestine and the dialectics of legal repression. Social Problems, 47, 445-472.

49.

Ron

(2000b). Territoriality and plausible deniability: Serbian paramilitaries in the Bosnian War. In Campbell

Brenner

(Eds.), Death squads in global perspective: Murder with deniability (pp. 287-312). New York, NY: St. Martin’s.

50.

Ron

(2003). Frontiers and ghettos: State violence in Serbia and Israel. Berkeley: University of California Press.

51.

Tenbrunsel

A. E.

Messick

(1999). Sanctioning systems, decision frames, and cooperation. Administrative Science Quarterly, 44, 684-707.

52.

Thornberry

Krohn

M. D.

(2000). The self-report method for measuring delinquency and crime. In Duffee

(Ed.), Criminal justice (pp. 33-84). Washington, DC: The National Institute of Justice.

53.

Trevino

(1992). The social effects of punishment in organizations: A justice perspective. Academy of Management Review, 17, 647-676.

54.

Trevino

Weaver

Reynolds

(2006). Behavioral ethics in organizations: A review. Journal of Management, 32, 951-990.

55.

Tsoref

Ayala

. (2010, July 7.) “Israel ranks 12th in percentage of population using Facebook.” Retrieved from http://www.haaretz.com/print-edition/business/israel-ranks-12th-in-percentage-of-population-using-facebook-1.300479

56.

Valentino

Huth

Balch-Lindsay

(2004). “Draining the sea”: Mass killing and Guerrilla warfare. International Organization, 58, 375-407.

57.

Watters

Biernacki

(1989). Targeted sampling: Options for the study of hidden populations. Social Problems, 36, 416-430.

58.

Weinstein

(2007). Inside rebellion: The politics of insurgent violence. Cambridge, UK: Cambridge University Press.

59.

Wood

E. J.

(2003). Insurgent collective action and civil war in El Salvador. Cambridge, UK: Cambridge University Press.

60.

Wood

E. J.

(2006). Variation in sexual violence during war. Politics & Society, 34, 307-341.

61.

Wood

E. J.

(2009). Armed groups and sexual violence: When is wartime rape rare? Politics & Society, 27, 131-162.

62.

Yehoshua

(2007, December 16). Abuses are revealed. Yedioth Aharonot (in Hebrew). p. 4.

63.

Yesh

Din

(2008). Investigation of criminal offenses of IDF soldiers against Palestinians and their property: Unit rankings 2006, 2007 (in Hebrew). Tel Aviv, Israel. Retrieved from http://www.yesh-din.org/userfiles/file/datasheets/%5BICAP%5D%20DS2%20Unit%20Index%20%5BHeb%5D.pdf